Application of pattern recognition techniques in ultrasound tissue characterization

An approach to the application of pattern recognition techniques is described by presenting examples of the tissue characterization from an ultrasound image. A Bayesian classifier is used to determine the feasibility of classifying tissue types and determining some properties of various feature spaces. An application of an unsupervised clustering technique that leads to more inherent properties of the features is described. A hypothesis-testing criterion is described and used to detect focal disease. The results to date are judged promising for tissue characterization of the liver and other abdominal organs